A retail company loads daily retail order CSV files into a single S3 path and uses Amazon Redshift Spectrum to query subsets of the more-than-100-column data. The third-party source produces over 30 CSV files per day (50–70 MB each). Users aggregate daily metrics, but query performance has degraded. Which combination of actions will resolve the performance issues with the least development effort? (Choose two.)

The correct actions are to configure the third-party application to produce files in a columnar format and to partition the order data in the S3 bucket by order date. Redshift Spectrum performs significantly better when querying columnar formats like Parquet or ORC because it can skip reading irrelevant columns, reducing the amount of data scanned. Partitioning by order date allows Redshift Spectrum to prune partitions and only scan data relevant to the daily aggregations, drastically improving query performance. Combining these two strategies addresses both the data format and data organization aspects, leading to substantial performance gains with minimal development effort. Developing an AWS Glue ETL job to combine files would add complexity and processing overhead without addressing the fundamental issues of row-based format or lack of partitioning. Configuring the third-party application to produce JSON format would worsen performance, as JSON is a semi-structured, often row-based format that is less efficient for analytical queries than columnar formats. Loading JSON into a SUPER type column in Redshift is not applicable here, as the problem describes Redshift Spectrum querying data directly from S3, not loading into a Redshift table.